rgit/src/git_fs/index.rs

use anyhow::Result;
use anyhow::bail;
use sha1::{Digest, Sha1};
use std::fs;
use std::{ffi::CStr, fmt::Display, fs::File, io::prelude::*, os::unix::fs::PermissionsExt};

use crate::git_fs::get_git_root;

#[derive(Clone, Debug, PartialEq)]
pub enum ObjectType {
    Regular,
    SymLink,
    GitLink,
}

impl Display for ObjectType {
    fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
        match &self {
            Self::Regular => write!(f, "Regular"),
            Self::SymLink => write!(f, "Symlink"),
            Self::GitLink => write!(f, "Gitlink"),
        }
    }
}

#[derive(Clone, Debug, PartialEq)]
pub struct IndexEntry {
    object_type: ObjectType,
    permissions: u16,
    hash: [u8; 20],
    name: String,
}

impl IndexEntry {
    fn from_bytes(bytes: &[u8]) -> Result<(Self, usize)> {
        let mode = &bytes[24..28];
        let ot_bin = mode[2] >> 4;
        let object_type = match ot_bin {
            0b1000 => ObjectType::Regular,
            0b1010 => ObjectType::SymLink,
            0b1110 => ObjectType::GitLink,
            _ => bail!("Invalid object type: {}", ot_bin),
        };
        let permissions = u16::from_be_bytes(mode[2..4].try_into()?) & !0xFE00;
        match (permissions, &object_type) {
            (0, ObjectType::GitLink) => (),
            (0, ObjectType::SymLink) => (),
            (0o755, ObjectType::Regular) => (),
            (0o644, ObjectType::Regular) => (),
            _ => bail!(
                "Invalid permissions (0o{:o}) for type {}",
                permissions,
                object_type
            ),
        };

        let hash: [u8; 20] = bytes[40..60].try_into()?;
        let cname = CStr::from_bytes_until_nul(&bytes[62..])?;
        let name = String::from(cname.to_str()?);

        let entry_size = usize::div_ceil(62 + cname.count_bytes() + 1, 8) * 8;

        Ok((
            IndexEntry {
                object_type,
                permissions,
                hash,
                name,
            },
            entry_size,
        ))
    }

    fn to_bytes(&self) -> Result<Vec<u8>> {
        let mut bytes = Vec::new();

        // ctime
        bytes.extend([0, 0, 0, 0]);
        // ctime nano
        bytes.extend([0, 0, 0, 0]);
        // mtime
        bytes.extend([0, 0, 0, 0]);
        // mtime nano
        bytes.extend([0, 0, 0, 0]);
        // dev
        bytes.extend([0, 0, 0, 0]);
        // ino
        bytes.extend([0, 0, 0, 0]);

        // mode
        let ot_bin: u16 = match self.object_type {
            ObjectType::Regular => 0b1000,
            ObjectType::SymLink => 0b1010,
            ObjectType::GitLink => 0b1110,
        };
        let perms: u16 = self.permissions | (ot_bin << 12);

        bytes.extend([0, 0]);
        bytes.extend(perms.to_be_bytes());

        // uid
        bytes.extend([0, 0, 0, 0]);
        // gid
        bytes.extend([0, 0, 0, 0]);
        // file size
        bytes.extend([0, 0, 0, 0]);
        // object name
        bytes.extend(self.hash);
        // flags
        let flags = u16::min(0xFFF, self.name.len() as u16);
        bytes.extend(flags.to_be_bytes());
        // entry path name
        bytes.extend(self.name.as_bytes());

        let padding = ((bytes.len() + 1).div_ceil(8) * 8) - bytes.len();
        bytes.extend(vec![0; padding]);

        Ok(bytes)
    }
}

#[derive(Debug, PartialEq)]
pub struct Index {
    version: u32,
    entries: Vec<IndexEntry>,
}

impl Default for Index {
    fn default() -> Self {
        Index {
            version: 2,
            entries: Vec::new(),
        }
    }
}

impl Index {
    pub fn load() -> Result<Self> {
        let path = get_git_root()?.join("index");

        if !path.exists() {
            return Ok(Index {
                version: 2u32,
                entries: Vec::new(),
            });
        }

        let mut file = File::open(&path)?;
        let mut content: Vec<u8> = Vec::new();
        file.read_to_end(&mut content)?;

        Index::from_bytes(content)
    }

    pub fn save(&self) -> Result<()> {
        let path = get_git_root()?.join("index");
        let mut file = File::create(&path)?;
        file.write_all(&self.to_bytes()?)?;
        Ok(())
    }

    pub fn add_file(&mut self, name: String, hash: [u8; 20]) -> Result<()> {
        let metadata = fs::metadata(&name)?;
        if metadata.is_dir() {
            bail!("Cannot add a directory to index")
        };

        let object_type = if metadata.is_symlink() {
            ObjectType::SymLink
        } else {
            ObjectType::Regular
        };

        let permissions = (metadata.permissions().mode() as u16) & !0xFE00;
        let entry = IndexEntry {
            name,
            hash,
            object_type,
            permissions,
        };

        self.insert_entry(entry);

        Ok(())
    }

    pub fn remove_file(&mut self, name: String) -> Result<()> {
        let entry = IndexEntry {
            name: name.clone(),
            hash: [0u8; 20],
            object_type: ObjectType::Regular,
            permissions: 0,
        };
        match self
            .entries
            .binary_search_by_key(&entry.name, |a| a.name.clone())
        {
            Ok(pos) => self.entries.remove(pos),
            Err(_) => bail!("No file {} in index", name),
        };

        Ok(())
    }

    fn insert_entry(&mut self, entry: IndexEntry) {
        match self
            .entries
            .binary_search_by_key(&entry.name, |e| e.name.clone())
        {
            Ok(pos) => self.entries[pos] = entry,
            Err(pos) => self.entries.insert(pos, entry),
        }
    }

    fn from_bytes(bytes: Vec<u8>) -> Result<Self> {
        match &bytes[0..4] {
            b"DIRC" => (),
            _ => bail!("Invalid index signature"),
        };

        let version = u32::from_be_bytes(bytes[4..8].try_into()?);
        let count = u32::from_be_bytes(bytes[8..12].try_into()?);

        let mut entries: Vec<IndexEntry> = Vec::with_capacity(usize::try_from(count)?);

        let mut offset = 12;

        for _i in 0..count {
            let (entry, entry_size) = IndexEntry::from_bytes(&bytes[offset..])?;
            offset += entry_size;
            entries.push(entry);
        }

        let mut hasher = Sha1::new();
        hasher.update(&bytes[..offset]);
        let hash = hasher.finalize();

        if bytes[offset..] != *hash {
            bail!("Index does not match its checksum")
        }
        Ok(Index { version, entries })
    }

    fn to_bytes(&self) -> Result<Vec<u8>> {
        let mut bytes: Vec<u8> = Vec::new();
        bytes.extend("DIRC".as_bytes());
        bytes.extend(self.version.to_be_bytes());
        let count: u32 = self.entries.len() as u32;
        bytes.extend(count.to_be_bytes());

        for entry in &self.entries {
            bytes.extend(entry.to_bytes()?);
        }

        let mut hasher = Sha1::new();
        hasher.update(&bytes);
        let hash = hasher.finalize();

        bytes.extend(hash);

        Ok(bytes)
    }
}

#[cfg(test)]
mod tests {
    use super::*;

    #[test]
    fn index_entry_to_bytes() {
        let entry1 = IndexEntry {
            object_type: ObjectType::Regular,
            permissions: 0o644u16,
            hash: [0x19u8; 20],
            name: String::from("src/git_fs/head.rs"),
        };
        let expected1: Vec<u8> = vec![
            0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
            0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x81, 0xa4,
            0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x19, 0x19,
            0x19, 0x19, 0x19, 0x19, 0x19, 0x19, 0x19, 0x19, 0x19, 0x19, 0x19, 0x19, 0x19, 0x19,
            0x19, 0x19, 0x19, 0x19, 0x00, 0x12, 0x73, 0x72, 0x63, 0x2f, 0x67, 0x69, 0x74, 0x5f,
            0x66, 0x73, 0x2f, 0x68, 0x65, 0x61, 0x64, 0x2e, 0x72, 0x73, 0x00, 0x00, 0x00, 0x00,
            0x00, 0x00, 0x00, 0x00,
        ];

        let entry2 = IndexEntry {
            object_type: ObjectType::GitLink,
            permissions: 0,
            hash: [0x19u8; 20],
            name: String::from("src/git_fs/head.r"),
        };
        let expected2: Vec<u8> = vec![
            0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
            0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0xE0, 0x00,
            0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x19, 0x19,
            0x19, 0x19, 0x19, 0x19, 0x19, 0x19, 0x19, 0x19, 0x19, 0x19, 0x19, 0x19, 0x19, 0x19,
            0x19, 0x19, 0x19, 0x19, 0x00, 0x11, 0x73, 0x72, 0x63, 0x2f, 0x67, 0x69, 0x74, 0x5f,
            0x66, 0x73, 0x2f, 0x68, 0x65, 0x61, 0x64, 0x2e, 0x72, 0x00,
        ];

        let entry3 = IndexEntry {
            object_type: ObjectType::SymLink,
            permissions: 0,
            hash: [0x19u8; 20],
            name: String::from("src/git_fs/head.rst"),
        };
        let expected3: Vec<u8> = vec![
            0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
            0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0xA0, 0x00,
            0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x19, 0x19,
            0x19, 0x19, 0x19, 0x19, 0x19, 0x19, 0x19, 0x19, 0x19, 0x19, 0x19, 0x19, 0x19, 0x19,
            0x19, 0x19, 0x19, 0x19, 0x00, 0x13, 0x73, 0x72, 0x63, 0x2f, 0x67, 0x69, 0x74, 0x5f,
            0x66, 0x73, 0x2f, 0x68, 0x65, 0x61, 0x64, 0x2e, 0x72, 0x73, 0x74, 0x00, 0x00, 0x00,
            0x00, 0x00, 0x00, 0x00,
        ];

        assert_eq!(entry1.to_bytes().unwrap(), expected1);
        assert_eq!(entry2.to_bytes().unwrap(), expected2);
        assert_eq!(entry3.to_bytes().unwrap(), expected3);
    }

    #[test]
    fn index_entry_from_bytes() {
        let expected1 = IndexEntry {
            object_type: ObjectType::Regular,
            permissions: 0o644u16,
            hash: [0x19u8; 20],
            name: String::from("src/git_fs/head.rs"),
        };
        let (entry1, len1) = IndexEntry::from_bytes(&vec![
            0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
            0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x81, 0xa4,
            0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x19, 0x19,
            0x19, 0x19, 0x19, 0x19, 0x19, 0x19, 0x19, 0x19, 0x19, 0x19, 0x19, 0x19, 0x19, 0x19,
            0x19, 0x19, 0x19, 0x19, 0x00, 0x12, 0x73, 0x72, 0x63, 0x2f, 0x67, 0x69, 0x74, 0x5f,
            0x66, 0x73, 0x2f, 0x68, 0x65, 0x61, 0x64, 0x2e, 0x72, 0x73, 0x00, 0x00, 0x00, 0x00,
            0x00, 0x00, 0x00, 0x00,
        ])
        .unwrap();

        let expected2 = IndexEntry {
            object_type: ObjectType::GitLink,
            permissions: 0,
            hash: [0x19u8; 20],
            name: String::from("src/git_fs/head.r"),
        };
        let (entry2, len2) = IndexEntry::from_bytes(&vec![
            0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
            0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0xE0, 0x00,
            0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x19, 0x19,
            0x19, 0x19, 0x19, 0x19, 0x19, 0x19, 0x19, 0x19, 0x19, 0x19, 0x19, 0x19, 0x19, 0x19,
            0x19, 0x19, 0x19, 0x19, 0x00, 0x11, 0x73, 0x72, 0x63, 0x2f, 0x67, 0x69, 0x74, 0x5f,
            0x66, 0x73, 0x2f, 0x68, 0x65, 0x61, 0x64, 0x2e, 0x72, 0x00,
        ])
        .unwrap();

        let expected3 = IndexEntry {
            object_type: ObjectType::SymLink,
            permissions: 0,
            hash: [0x19u8; 20],
            name: String::from("src/git_fs/head.rst"),
        };
        let (entry3, len3) = IndexEntry::from_bytes(&vec![
            0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
            0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0xA0, 0x00,
            0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x19, 0x19,
            0x19, 0x19, 0x19, 0x19, 0x19, 0x19, 0x19, 0x19, 0x19, 0x19, 0x19, 0x19, 0x19, 0x19,
            0x19, 0x19, 0x19, 0x19, 0x00, 0x13, 0x73, 0x72, 0x63, 0x2f, 0x67, 0x69, 0x74, 0x5f,
            0x66, 0x73, 0x2f, 0x68, 0x65, 0x61, 0x64, 0x2e, 0x72, 0x73, 0x74, 0x00, 0x00, 0x00,
            0x00, 0x00, 0x00, 0x00,
        ])
        .unwrap();

        assert_eq!(entry1, expected1);
        assert_eq!(entry2, expected2);
        assert_eq!(entry3, expected3);

        assert_eq!(len1, 88);
        assert_eq!(len2, 80);
        assert_eq!(len3, 88);
    }

    #[test]
    fn index_insert_entry() {
        let mut index = Index::default();

        let entry1 = IndexEntry {
            object_type: ObjectType::Regular,
            permissions: 0o644u16,
            hash: [0x19u8; 20],
            name: String::from("a"),
        };
        let entry2 = IndexEntry {
            object_type: ObjectType::Regular,
            permissions: 0o644u16,
            hash: [0x19u8; 20],
            name: String::from("b"),
        };
        let entry3 = IndexEntry {
            object_type: ObjectType::SymLink,
            permissions: 0,
            hash: [0x19u8; 20],
            name: String::from(".hello"),
        };
        let entry4 = IndexEntry {
            object_type: ObjectType::Regular,
            permissions: 0o644u16,
            hash: [0x20u8; 20],
            name: String::from("b"),
        };

        index.insert_entry(entry1.clone());
        index.insert_entry(entry2.clone());
        index.insert_entry(entry3.clone());
        index.insert_entry(entry4.clone());

        let expected = Index {
            entries: vec![entry3, entry1, entry4],
            ..Default::default()
        };

        assert_eq!(index, expected);
    }

    #[test]
    fn index_remove_entry() {
        let entry1 = IndexEntry {
            object_type: ObjectType::Regular,
            permissions: 0o644u16,
            hash: [0x19u8; 20],
            name: String::from("a"),
        };
        let entry2 = IndexEntry {
            object_type: ObjectType::Regular,
            permissions: 0o644u16,
            hash: [0x19u8; 20],
            name: String::from("b"),
        };
        let entry3 = IndexEntry {
            object_type: ObjectType::SymLink,
            permissions: 0,
            hash: [0x19u8; 20],
            name: String::from(".hello"),
        };

        let mut index = Index {
            entries: vec![entry3.clone(), entry1, entry2.clone()],
            ..Default::default()
        };

        index.remove_file(String::from("a")).unwrap();

        let expected = Index {
            entries: vec![entry3, entry2],
            ..Default::default()
        };

        assert_eq!(index, expected);
    }
}